Browser-native vocal separation

Acapella Extractor

CPU (WASM fallback)

Source

Upload an audio file to separate.

Empty

Drop a song here

MP3, WAV, M4A, FLAC, AAC, OGG

Choose audio

Hybrid laptops: pick Discrete GPU for NVIDIA.

WebGPU unavailable here — extraction uses CPU (WASM).

Speed options

Download once — cached in this browser.

Needed

Hugging Face · tap ↓ to cache locally

MDX fine-tune with optional spectral denoise for noisy mixes

Upload a track to begin.

Your waveforms will appear here

Upload a track, choose a model, then hit Extract in the header above.